New Understanding of Life and Its Processes: Structure of Proteins and DNA

spacer spacer

Photo: Pein Tang Pei Tang, University of Pittsburgh School of Medicine.


Research in general anesthesia helps us understand self-awareness, maybe the essential trait of being human.

f you've had surgery, you may recall noticing the brightness of the ceiling lights in the operating room. Next thing you know your eyes peel open and a voice asks, "How are you feeling?" In the meantime, someone has sliced your body open and put it back together.

General anesthesia is one of the wonders of medicine. The ability to induce a deeply unconscious, immobile state makes possible life-saving procedures no one could have imagined 150 years ago, when surgeons were "saw-bones" and tooth extraction or amputation of a limb was excruciating beyond description. The discovery of ether's anesthetizing effect in the 1840s has led in recent decades to new, safer anesthetics. An estimated 15 million Americans undergo general anesthesia annually. Despite the advances in pharmacology and wide use, however, no one can tell you how anesthetics do what they do.

"General anesthesia has been used for more than a century," says Pei Tang, a physical chemist and assistant professor of anesthesiology and pharmacology at the University of Pittsburgh School of Medicine (UPSM). "It's one of the most important tools of medicine, and yet it remains mysterious. Despite years of research, we still don't understand the molecular mechanism."

It's an alluring mystery. What's known about general anesthetics suggests they are a peculiar class of drugs, with their own way of acting in the body that doesn't fit the standard model of how drugs work. Understanding the molecular details would likely lead to better anesthetics, with fewer side effects, but perhaps even more importantly it should tell us something about consciousness itself, one of the grandly intriguing questions in science.

An unconsciousness deeper than sleep, general anesthesia involves complete muscle relaxation — the loss of fight or flight response, which persists in sleep — and total amnesia. Arguably the ultimate altered state, it's like flipping a switch to temporarily turn off the nervous system. Understanding how this happens will increase our feeble knowledge of what it means biochemically to be switched "on" with the self-awareness that may be the essential trait of being human.

Tang is part of a UPSM team that uses a range of techniques to investigate how anesthetics work. In collaboration with Pittsburgh Supercomputing Center scientists Marcela Madrid and Troy Wymore, she has used computational methods to simulate how the drugs interact with the cellular membranes where they have their effect. The results of these studies challenge accepted thinking and offer support for an emerging new hypothesis.

Final structure of the DMPC membrane in the fluid phase. The membrane consists of 200 DMPC lipid molecules — phosphate head groups (red spheres) and hydrocarbon chains (blue) — surrounded by 5,483 water molecules (yellow). The insert shows the structure at the start of the simulation.

Download a larger version of this image.

A Tale of Two Theories

In the normal, undrugged state of consciousness, sense perceptions trigger a chain of events that releases electrical signals, in the form of ion flow — sodium, calcium, potassium and other ions — through channels in the cell walls, better known to biologists as membranes. General anesthetics appear to exert their effect by changing ion flow through these membrane channels, either slowing it down or speeding it up. Thinking about how this happens falls generally into two schools: the lipid theory and the protein theory.

The lipid theory, the older point of view, says that general anesthetics work by their ability to dissolve in lipids, the fat that forms the cell membrane and seals it against the watery environs inside and outside the cell. This view arose, initially, from finding that the potency of an anesthetic chemical corresponds to how well it dissolves in olive oil. Following from this awareness, some experiments suggest that anesthetics make the lipid more fluid, a structural loosening that relaxes and changes the shape of the channels that control ion flow.

The protein theory is more recent and says, on the other hand, that general anesthetics interact directly with the channels, which are complex proteins, rather than indirectly through the lipids. This idea is closer to the lock-and-key model of how most drugs work, by binding with a specific site on a protein and blocking its interaction with other molecules. This model doesn't fit well with anesthetics, however, because there's little structural similarity among different anesthetics, one reason the lipid theory gains adherents. The drug can't have its effect by acting like a key in a lock, in other words, when there are so many different shapes that work.

Experiments have shown, nevertheless, that a range of anesthetics have the ability to radically slow down flashes of light from a protein called luciferase, which makes the tails of fireflies glow. Luciferase doesn't exist in cell membranes, so in this case at least, the effect is direct. Still the question is how. Anesthetics are "low-affinity" drugs — they must be administered in relatively high concentration, another indication that they don't latch on tightly at a specific site like most other drugs, which are high affinity. In experiments with membranes, as opposed to isolated proteins, it's extremely difficult to test the protein theory because it's nearly impossible to disentangle direct from indirect action.

For that, computational simulations, which track the atom-by-atom details of the drug-membrane-channel interactions, have the potential to break through the theoretical logjam. "Only new techniques like large-scale simulations," says Tang, "that permit analysis at or near atomic resolution can test these theories."


Electrostatic potential, the quantity of electric charge, is shown mapped onto the electron-density surface for halothane (left) and sevoflurane (right). Color (increasing from blue to red) indicates charge. The skeleton of molecular structure is also visible.

Download a larger version of this image.

Lipids, Drugs & Protein in a Water Sandwich

Cellular membranes are complex molecular assemblies involving tens of thousands of atoms, and only in recent years has computational capability evolved to make it feasible to simulate these structures. Starting in 1999, as the first step in a staged process, Tang constructed and tested a computational model of a cellular membrane called DMPC (short for dimyristoylphosphatidylcholine).

Lipids are long-chain molecules with one end, the "head-group," that gets along chemically with water and a tail-end that's hydrophobic, water avoiding. Cellular membranes are bilayers of these molecules, with head-groups facing outward and inward to meet with water while the tails converge in the hydrophobic interior. Tang built a bilayer of 200 DMPC molecules and added a layer of water on each side, 5,483 water molecules, to realistically represent the chemical environment. Using the NAMD2 molecular dynamics program on PSC's CRAY T3E, she simulated how this structure, a lipid bilayer in a water sandwich, changes under realistic conditions. The results gave structural parameters for the membrane that agreed well with experimental data.

The next step was to compute the structural details and electronic properties of two frequently used general anesthetics, halothane and sevoflurane. These quantum-theory calculations provide information such as the precise distance and angles between atoms in the drug molecules along with the electronic charges that influence their interaction with other molecules. The results, which again agreed well with experimental data, provided parameters for Tang's next step: to simulate how anesthetic drugs act within the membrane-channel environment.

With the groundwork in place, Tang used NAMD2 and the CRAY T3E for a simulation that included 10 halothane molecules inside the DMPC membrane, which itself included a protein molecule, gramicidin, as an ion channel through the lipid bilayer. This very large-scale computation included 38,724 atoms and tracked the molecular movements for two nanoseconds (a billionth of a second) with a freeze-frame picture of the system every femtosecond (a millionth of a billionth of a second).

Halothane in membrane with channel. In this image from the simulation, ten halothane molecules (clustered balls) interact with gramicidin (blue coil), a protein inserted in the DMPC membrane as an ion channel. The simulation shows that halothane molecules move toward the entry of the channel.

Download a larger version of this image.

After about 240 hours of processing time (on 128 processors), the results show, contrary to accepted understanding, that halothane in the center of the lipid moves away from this hydrophobic environment toward the water. "People have believed that all anesthetics prefer hydrophobic environments," says Tang. "Even those who support the protein theory thought that anesthetics interacted with hydrophobic cavities in the channel. We found, however, that the halothane always moves away from the center toward the interface between the lipids and water."

This result supports an emerging hypothesis that bridges between the two competing theories and suggests that general anesthetics act at the channel-water interface. The simulation lends further support to this view by showing that halothane interacts strongly with one of the protein's amino-acid side-chain (tryptophan), which resides close to the channel. This finding appears to coincide with recent UPSM laboratory work.

To further test the interface hypothesis, Tang looks forward to PSC's Terascale Computing System. She'd like to simulate other anesthetics and extend simulation time into the millisecond range. She also plans to simulate compounds structurally very similar to anesthetics but that produce no anesthetic effect. These studies, she expects, will help pinpoint the molecular events that lead to general anesthesia. "With these simulations," says Tang, "I believe we'll be able to draw some conclusions that will lead us closer to solving this mystery."

Download PDF The Road to La-La Land
from Projects in Scientific Computing 2001
Researchers Pei Tang, University of Pittsburgh
Hardware CRAY T3E
Software NAMD2
Related Material
on the Web
Department of Anesthesiology & Critical Care Medicine, University of Pittsburgh
References Pei Tang, Igor Zubryzcki & Yan Xu, "Ab Initio Calculation of Structures and Properties of Halogenated General Anesthetics: Halothane and Sevoflurane," Journal of Computational Chemistry, 22, 436-44 (2001).

Igor Z. Zubryzcki, Yan Xu, Marcela Madrid & Pei Tang, "Molecular dynamics simulations of a fully hydrated dimyristoylphosphatidylcholine membrane in liquid-crystalline phase," Journal of Chemical Physics, 112, 3437-41 (2000).

Pei Tang and Yan Xu, " Large-scale molecular dynamics simulations of general anesthetic effects on the ion channel in the fully hydrated membrane: The implication of molecular mechanisms of general anesthesia," Proceedings of the National Academy of Sciences 99 (25), 16035-16040 (2002).
Writing: Michael Schneider
HTML Layout/Coding: R. Sean Fulton
Graphic: PSC Logo © Pittsburgh Supercomputing Center (PSC)
Revised: July 27, 2001